zxs / tungsten-replicator

Automatically exported from code.google.com/p/tungsten-replicator
0 stars 0 forks source link

Star - New slaves fail to synchronize after a while of operation #335

Open GoogleCodeExporter opened 9 years ago

GoogleCodeExporter commented 9 years ago
What steps will reproduce the problem?

1. Set up a star cluster and run operations as normal
2. After an indeterminate time, when new slaves are added, they will fail with 
the pendingError "Stage task failed: remote-to-thl"

What is the expected output?
When:
1) A star cluster has been configured
2) Endpoints can be added without issue
3) No changes have occurred other than normal database operations

Then adding a new endpoint via the exact same method should succeed.

What do you see instead?
After an indeterminate time, the hub-to-endpoint master-slave relationship 
fails, and the new endpoint's slave service fails to synchronize with the hub.

What version of the product are you using?
2.0.6-584

On what operating system?
Linux (Ubuntu 11.10, 64-bit)

Please provide any additional information below.
The logs on the master cycle with the following for a while before the slave 
gives up:
INFO   | jvm 1    | 2012/05/21 10:33:09 | 
com.continuent.tungsten.replicator.thl.log.LogPositionException: Log seek 
failure: expected seqno=0 found seqno=3453
INFO   | jvm 1    | 2012/05/21 10:33:09 |   at 
com.continuent.tungsten.replicator.thl.log.LogConnection.next(LogConnection.java
:604)
INFO   | jvm 1    | 2012/05/21 10:33:09 |   at 
com.continuent.tungsten.replicator.thl.ConnectorHandler.run(ConnectorHandler.jav
a:331)
INFO   | jvm 1    | 2012/05/21 10:33:09 |   at 
java.lang.Thread.run(Thread.java:662)
INFO   | jvm 1    | 2012/05/21 10:33:09 | 2012-05-21 10:33:09,496 [central - 
connector-handler-ecsyncvmstation] INFO  thl.ConnectorHandler Terminating THL 
client connection from source ID: ecsyncvmstation
INFO   | jvm 1    | 2012/05/21 10:33:09 | 2012-05-21 10:33:09,727 [central - 
ConnectorHandler: initializing] INFO  thl.Protocol THL protocol buffering 
enabled: size=10
INFO   | jvm 1    | 2012/05/21 10:33:09 | 2012-05-21 10:33:09,783 [central - 
ConnectorHandler: initializing] INFO  thl.ConnectorHandler New THL client 
connection: sourceID=ecsyncvmstation heartbeatMillis=3000
INFO   | jvm 1    | 2012/05/21 10:33:09 | 2012-05-21 10:33:09,783 [central - 
ConnectorHandler: initializing] INFO  thl.ConnectorHandler Client log checking 
disabled; not checking for diverging histories
INFO   | jvm 1    | 2012/05/21 10:33:09 | 2012-05-21 10:33:09,808 [central - 
connector-handler-ecsyncvmstation] ERROR thl.ConnectorHandler Connector handler 
terminated by THL exception: Log seek failure: expected seqno=0 found seqno=3453

Notes:
In looking for duplicates, I considered issue #281, but I am unsure if this is 
the same thing.  Please inform me of any additional information you require.

Original issue reported on code.google.com by aeon.des...@gmail.com on 21 May 2012 at 2:58

GoogleCodeExporter commented 9 years ago
This problem happens even with 2.1.0 as well. No matter how many clean installs 
you do it still says Connector handler terminated by THL exception: Log seek 
failure: expected seqno=0 found seqno=166981. Any workarounds or should we wait 
for 2.1.2?

Original comment by sastryRa...@gmail.com on 8 Jul 2013 at 4:38

GoogleCodeExporter commented 9 years ago
installation of star clusters have changed dramatically in the latest version, 
and that could affect the cluster behavior.
Could you share the details of how the installation was done?

Original comment by g.maxia on 8 Jul 2013 at 4:41